I would like to express my gratitude to the staff of Lanzhou University’s Centre for Studies of Ethnic Groups in Northwest China, and my supervisor, Professor Wu Mu 武沐, in particular, for generously facilitating my research in Jishishan County. I would also like to thank the anonymous reviewers for their highly beneficial comments and suggestions, Charlotte Thionois and Isabelle Charleux for their kind and detailed proof reading, and Katia Buffetrille for her warm encouragement all the way through the submission and review process. Soli Deo Gloria.

Introduction

2While various scholars have conducted linguistic and sociolinguistic analyses of both Bonan varieties (see below), a broad sociolinguistic assessment of language vitality across the Gansu Bonan-speaking villages has not yet been produced. This article, based on survey work in 2012 and 2013 as well as on subsequent ethnographic fieldwork, seeks to fill that gap, providing a general overview of the status of the language in Gansu as a basis for further work by linguists, other scholars or policymakers.

Theoretical background

Measuring language endangerment

3There have been a number of scales used to assess language vitality or endangerment, of which four of the most important ones are Fishman’s (1991) Graded Intergenerational Disruption Scale (GIDS), UNESCO’s language vitality assessment framework (Brenzinger et al. 2003), Ethnologue’s Expanded Graded Intergenerational Disruption Scale (EGIDS) (Lewis & Simons 2010) and the Catalogue of Endangered Languages (ELCat)’s Language Endangerment Index (LEI) (Lee & Van Way 2016). GIDS was one of the pioneering scales, an eight levels “Richter”-like gradation emphasising intergenerational transmission and domains of use as the important factors in language vitality. Lewis and Simons built their model on GIDS, expanding Fishman’s scale to thirteen levels, and basing their assessment of a language on five key factors (see tables 1, 3). The UNESCO framework sought to move beyond the use of a single “score” to assess a language’s vitality by making use of different scales for nine influencing factors, stating that “none of these factors should be used alone” (Brenzinger et al. 2003, p. 7, emphasis original; see table 3). However, in practice the “Degree of Endangerment” scale shown on UNESCO’s World Atlas of Languages website is solely based on the first factor, “Intergenerational Language Transmission” (Moseley 2010, pp. 11-12); Lewis and Simons regard this factor as the most salient of UNESCO’s nine (Lewis & Simons 2010, p. 107) and use it for direct comparison with their EGIDS categories (see table 1). Lee & Van Way agree with Brenzinger et al. that multiple factors should not be lumped together into one, and yet critique the UNESCO framework for not being designed in such a way that an overall score can be ascertained for comparison with other languages. They also highlight a problem common to all attempts at scaling language vitality, namely the lack of detailed information on a large portion of the world’s languages (Lee & Van Way 2016, pp. 276-277). Taking these issues into account, ELCat’s LEI model bases a language’s assessment on up to four factors (intergenerational transmission, absolute number of speakers, speaker number trends, and domains of use), aggregating them into one score, while adding an additional “certainty” scale depending on which and how many factors are known with sufficient confidence to be used in the calculation (Lee & Van Way 2016, pp. 278-286; see tables 7-8 below). These latter three internationally used scales each contain a level for Bonan: EGIDS – 6b, Threatened (Eberhard et al. 2023); UNESCO – 3, Definitely Endangered (equivalent to EGIDS level 7 (Shifting) according to Lewis and Simons’ comparison) (UNESCO 2021); and LEI – Threatened at 40 percent certainty (Catalogue of Endangered Languages 2023a, 2023b). However, neither the Ethnologue nor UNESCO’s World Atlas of Languages differentiates between Gansu and Qinghai Bonan, and, for example, the sources on the ElCat website are around 20 years old. The following analysis seeks tentatively to confirm or update the results on these sites.

Table 1. Lewis and Simons’ comparison of their Expanded Graded Intergenerational Disruption Scale (EGIDS) with UNESCO’s Degree of Endangerment Scale (taken from Lewis & Simons 2010, p. 110, courtesy of the editorial board of Revue Roumaine de Linguistique)

Expanded Graded Intergenerational Disruption Scale
LEVEL	LABEL	DESCRIPTION	UNESCO
0	International	The language is used internationally for a broad range of functions.	Safe
1	National	The language is used in education, work, mass media, government at the nationwide level.	Safe
2	Regional	The language is used for local and regional mass media and governmental services.	Safe
3	Trade	The language is used for local and regional work by both insiders and outsiders.	Safe
4	Educational	Literacy in the language is being transmitted through a system of public education.	Safe
5	Written	The language is used orally by all generations and is effectively used in written form in parts of the community.	Safe
6a	Vigorous	The language is used orally by all generations and is being learned by children as their first language.	Safe
6b	Threatened	The language is used orally by all generations, but only some of the child-bearing generation are transmitting it to their children.	Vulnerable
7	Shifting	The child-bearing generation knows the language well enough to use it among themselves, but none are transmitting it to their children	Definitely Endangered
8a	Moribund	The only remaining active speakers of the language are members of the grandparent generation.	Severely Endangered
8b	Nearly Extinct	The only remaining speakers of the language are members of the grandparent generation or older who have little opportunity to use the language.	Critically Endangered
9	Dormant	The language serves as a reminder of heritage identity for an ethnic community. No one has more than symbolic proficiency.	Extinct
10	Extinct	No one retains a sense of ethnic identity associated with the language, even for symbolic purposes.	Extinct

4At the same time as these scales have been developed, scholars have questioned the reliance on quantitative analysis in the whole field of endangered language assessment and advocacy; for example, Dobrin et al. (2009, p. 41) claim that the comparison and ranking of languages according to official degrees of endangerment reduce the prioritisation of language documentation to a “numbers game”, while Hill (2002, pp. 127-128) argues that enumeration essentialises and individualises a language as a discrete unit, despite the acknowledged arbitrary nature of language and dialect classification (see also Mühlhäusler 1996, pp. 2-8). In her response to Lee and Van Way (2016), Grenoble (2016) highlights three problems with assessing language vitality through enumeration: commodification of the language (cf. Heller 2003; Dobrin et al. 2009), the challenges involved in classifying the number of speakers (cf. Moore et al. 2010; Jaffe 2007, p. 68; see also Tsunoda 2006, pp. 117-133 for an overview of various speaker classification proposals) and the general point of reducing “complex social dynamics to a single number or a set of numbers” (Grenoble 2016, pp. 293-294). She shows how “[a]ggregate measures assume a homogeneity across speech communities that does not exist”, citing examples from Russia where vitality varies from region to region, between urban and rural settings, and from village to village (ibid., p. 296). Other scholars are more positive towards the endangerment scales and quantitative approaches while not denying their limitations (Bradley & Bradley 2019, pp. 14-29; Gomashie & Terborg 2021; Bromham et al. 2020, Bromham et al. 2022). In the 2023 volume Endangered Languages in the 21^st Century (Derhemi & Moseley), which takes an optimistic view of language revitalisation efforts using examples from around the world, various contributors promote community-based approaches alongside the use of surveys and scales, in order to produce more holistic results and cooperative strategies for maintenance (see Drude et al. 2023, pp. 21-40; Lewis 2023, pp. 105-118; Bradley 2023, pp. 119-125; Haboud & Ortega 2023, pp. 284-302). In Western China specifically, Roche (2014) uses UNESCO’s nine factors to provide an overview of the outlook for cultural Tibet’s minority languages, and Roche (2014) and Roche & Suzuki (2018) make extensive use of EGIDS in assessing the vitality of the region’s individual languages. Dwyer (2011) assesses three Tibeto-Burman languages (Amdo Tibetan, Wutun and Sanie) according to both UNESCO’s nine factors and EGIDS, preferring the former over the latter, and arguing that such a broader evaluation enables the researcher to pinpoint and act on specific problematic areas. While the present study does produce “scores” according to the three scales, it also combines the quantitative analysis with qualitative data by utilising the various factors used in the three frameworks, analysing the variability across different villages, and including ethnographic data from field research, in order to produce a holistic picture of Gansu Bonan’s sociolinguistic status.

Sociolinguistic context

5The Bonan are situated in northwest China’s Amdo sprachbund, also known as the Qinghai-Gansu sprachbund or other variants, and which contains around fifteen different linguistic entities within four major language families: Sinitic, Bodic, Mongolic and Turkic (Dwyer 1995; Slater 2003, pp. 6-9, 2021; Janhunen 2007, pp. 85-87). Within such a cultural and linguistic milieu, the various languages have undergone profound mutual interaction, resulting in “Altaicization (of the local varieties of Chinese), Tibetanization (of the local varieties of Turkic and Mongolic, and some varieties of Chinese), as well as […] Sinicization (of some of the local varieties of Mongolic” (Janhunen 2015, p. 122). Specifically, the two Bonan varieties have developed separately according to their respective linguistic environments, with the result that the Qinghai variant shows strong contact effect with Amdo Tibetan (Xi 1986, p. 63; Fried 2010, p. 15), while the Gansu variant has received heavy Sinitic and Sinophone Islamic influences (Xi 1986, p. 63; Li & Dwyer 2020, pp. 3-4). At a broader level, Bonan can be classed as a minority language of cultural Tibet, which Roche & Suzuki define as all languages spoken in Tibetan regions that are “not ‘Tibetan’ or ‘Chinese’” (2018, p. 1238; see also Roche 2014). In this context, the two Bonan varieties face differing, yet related challenges. Qinghai Bonan, as an unrecognised language within a Tibetan prefecture, faces the encroachment of a “hegemonically imagined single Tibetan language” (Roche 2022, p. 15), where a drive for “purity” in Tibetan is not only promoted by state ethnic (Ch. minzu 民族) definitions, but supported by Tibetan elites at home and abroad in juxtaposition to Han Chinese dominance (Thurston 2018; Roche 2019b; see also Tso & Turin 2019). While the Gansu Bonan have the advantage of being an official minzu, and are located just outside the Tibetan hegemonic sphere, their language is in practice unsupported by the state, and due to the linguistic and cultural dominance of the regional Sinophone Muslims, they in fact also face the threat of “Huiification” (i.e. that of losing any distinctiveness from the wider Northwest Mandarin-speaking Hui Muslim community) in a similar way to how the Bonan Tu are facing “Tibetanisation” (Battye 2019).

6One factor contributing to the assimilatory pressures facing both communities is a lack of written or oral representation in their language. Neither the Qinghai nor the Gansu variety of Bonan has an official or unofficial orthography (Li & Dwyer 2020, p. 4), and in this respect too they face a somewhat parallel situation. In Qinghai, romanised scripts have been developed for Mongghul as well as Mangghuer Tu (Li 1988; Fried 2010, pp. 5-6; Slater 2003, p. 10; Limusishiden & Dede 2012, p. 106), but they have not, to my knowledge, been utilised in the Bonan Tu context. Likewise, a Latin orthography has been approved and utilised for the linguistically related Dongxiang ethnicity (Ch. Dongxiangzu 东乡族) (Slater 2003, p. 6), yet it has not been adapted for use among the Gansu Bonan. Due to the divergence of the two Bonan-speaking communities, both in terms of religion and of official ethnic classification, there has been no proposal for a comprehensive orthography for use in both variants. In terms of the media domain, all provincial and prefectural television stations broadcast in Mandarin (Jishishan), or Mandarin and Amdo Tibetan (Tongren) (Yefenghaiyun-Awei 2023; Qinghai sheng guangbo dianshiju chuanmei jigou guanlichu 2022), and internet and social media platforms are also dominated by Mandarin Chinese. In contrast, neither variety of Bonan has much opportunity for expansion into new forms of media, apart from the use of voice messaging or the uploading of recorded videos on social media, both of which would only reach a limited audience. In sum, such a lack of written and oral representation adds to the marginalisation of both varieties of Bonan, as they are crowded out in these domains by the regionally dominant written and broadcast languages of Mandarin Chinese and Amdo Tibetan.

Bonan language classification and previous research

7The 19^th-century Russian ethnographer Grigorij Potanin was the first to use the title “Shirongol” to describe the “settled Mongols in Amdo” as distinct from the Mongols of the plateau (Potanin [1893] 2014); linguists have subsequently classified Bonan, along with Santa (Dongxiang), Mongghul and Mangghuer as the “Shirongolic” complex of Southeastern Mongolic (Janhunen et al. 2007, p. 180; Fried 2010, p. 2; Li & Dwyer 2020, p. 4). B. X. Todaeva’s 1966 grammar of Bonan, based on fieldwork conducted as part of the Sino-Russian linguistic expedition (1955-1956), established the similarity between the two variants (Janhunen et al. 2007, p. 180), and in subsequent decades Chinese and international scholars have produced grammatical sketches and lexicons of both Qinghai and Gansu Bonan. In Qinghai, Chen Naixiong and Činggeltei (Ch. transcription Qingge’ertai)’s research in the 1980s produced a lexicon (Chen 1986), text materials (Chen 1987) and a comparison of Bonan and Mongolian (Chen & Qingge’ertai 1987); Xi (1986) also produced a phonetic and grammatical overview. In English, Chuluu (1994) published a summary of Chen 1986, Chen 1987 and Chen & Qingge’ertai 1987; Wu (2003) provided an overview and grammatical sketch, and Fried (2010) produced a grammar as his doctoral dissertation. Chen Naixiong’s later articles focused on Bonan more generally, including research on the Gansu variant and comparisons of the two (1990a, 1990b, 1995a, 1995b, 1995c).

8Based in Dahejia 大河家 for his field research, Todaeva focused primarily on Gansu Bonan; as well as his 1966 grammar, which included language materials and a lexicon, he also produced two smaller grammar sketches (1963, 1997). In the 1980s Buhe & Liu provided a grammatical sketch of Gansu Bonan based on fieldwork in Dadun Village (1982), and Charles Li published a number of English articles on language change and contact in the region, including the influence of Hezhou 河州 Chinese on Bonan (1986). Hua Kan (1992) also provided a summary of Tibetan loan words in Gansu Bonan. In recent years phonetic studies in Gansu have been undertaken by Ma (2015), Degejihu et al. (2014) and Degejihu et al. (2015), and three dictionaries have been produced: Mo & Zhang 2008 (Bonan-Chinese/English), Ma & Satō 2016 (Bonan-Chinese), and Li & Dwyer 2020 (Bonan-English/Chinese), the latter based on Li’s field research in the 1980s. This brief summary shows that a reasonable amount of research has been undertaken in both Tongren and Jishishan; however, Li and Dwyer still conclude that “both varieties of the Bonan language remain under-described” (2020, p. 5).

9There have been fewer sociolinguistic surveys conducted among the two communities. In Tongren, Wang Yuanxin conducted surveys in Nianduhu and Lower Bao’an villages (Wang 2009a, 2009b) concluding that though Bonan was relatively maintained within the home and village domains, language transmission was starting to weaken across generations. Roche and Lcag Mo Tshe Ring (2013) show how all religious rituals in Nianduhu Village are conducted in Tibetan, contributing to the marginalisation of Bonan, and Roche’s (2019a) sociolinguistic analysis confirms that language shift is occurring, although at different rates between villages. Through interviews conducted with individuals from various Jishishan villages, Jian Zhixiang concluded that Gansu Bonan was spoken relatively infrequently there by the younger generation, although the language was maintained to a greater degree in Ganhetan compared to other villages (2006, pp. 352-360). Bai’s (2019) cross-village survey highlighted the relative importance of Bonan-speaking mothers compared with fathers in transmitting the language to the next generation. Both based on research in Dadun, Ying (2006) described Bonan as lacking in vitality and Yang (2010, pp. 52-55) placed the language between level 3 (“Definitely Endangered”) and 2 (“Severely Endangered”) on the UNESCO scale. Satō Nobuharu (2006) summarised the sociolinguistic situation in the five villages which have historically had significant Bonan-speaking populations (see table 2); however, he did not provide village by village data to support his conclusions. The following survey builds on the sociolinguistic work already conducted for the Gansu Bonan, and, in particular, tests Satō’s conclusions against data from the five villages.

Table 2. Satō’s (2006, courtesy of Gansu renmin chubanshe) assessment of Gansu Bonan endangerment levels by village (author’s translation; for the original, see his article)

Danger Level	Bonan Village
1. Relative to other villages, the influence of the Chinese language is not as strong.	Dadun Village 大墩村
2. Strongly influenced by the Chinese language.	Ganhetan Village 甘河滩村
3. Most people do not speak Bonan, or speak very simple Bonan.	Gaoli Village 高李村
4. Bonan is not spoken.	Xiaojia Village 肖家村 Xietao Village 斜套村

Methodology

10Table 3 shows the factors utilised in the three global language endangerment frameworks which provide a level for Bonan, together with a combined summary list of ten factors, taking into account the overlaps between them. In order to update or confirm the levels given on the scales, the ten factors will be analysed, based on a sociolinguistic survey conducted across Satō’s five villages in 2012 and 2013, as well as ethnographic data gained from subsequent field research in the area.

Table 3. Factors utilised in the EGIDS, UNESCO and LEI’s respective endangerment frameworks (from Lewis & Simons 2010; Brenzinger et al. 2003; Lee & Van Way 2016), with a combined list of ten factors taking into account the overlaps

Factor	EGIDS	UNESCO	LEI	Combined
1	Identity function (domain)	Intergenerational language transmission	Intergenerational transmission	Intergenerational transmission
2	Level of Official Use	Absolute number of speakers	Absolute number of speakers	Absolute number of speakers
3	State of intergenerational language transmission	Proportion of speakers within the total population	Speaker number trends (whether increasing or decreasing)	Proportion of speakers
4	Literacy status	Trends in existing language domains	Domains of use	Speaker number trends
5	Youngest generation of proficient speakers	Response to new domains and media		Domains of use and trends
6		Materials for language education and literacy		Language attitudes
7		Governmental and institutional language attitudes and policies, including official status and use		Literary status
8		Community members’ attitudes towards their own language		Response to new domains and media
9		Amount and quality of documentation		Government policies/official status and use
10				Amount and quality of documentation

Table 4. Questions on language use, ability and attitudes in the author’s 2012-2013 Gansu Bonan language survey

1	In what situation do you speak Bonan?	Always	Usually with relations, friends and neighbours	Only at home	I generally don’t speak it
2	How would you describe your Bonan language ability?	Fluent	Relatively fluent	OK	I can speak a little	I can’t speak it
3	You believe the Bonan language will:	Quickly die out	Slowly die out	Continue to survive
4	As far as you’re concerned, protecting the Bonan language:	Is a very important mission	Is a relatively important mission	Is not very important	Is not important at all

Results by village

Ganhetan (population = 2 086; total number of interviewees = 62)

Education

14As an unsupported language with no orthography, Bonan is not utilised to any formal degree in Ganhetan primary school. In addition, due to the fact that the teachers are rotated in from other areas, they will not necessarily be Bonan speakers themselves and will thus use Northwest Mandarin for general communication with pupils. Finally, the official drive within the educational system to encourage standard Mandarin pronunciation (Ch. putonghua 普通话, “common speech”) further marginalises Bonan as a third-level language below standard and Northwest Mandarin. From Middle School upwards, the pupils will move to Dahejia Township or further afield, where they will mix predominantly with non Bonan speakers, only returning to the village at weekends and in the holidays; hence, from then on, communicating in their mother tongue will become a less regular feature of their everyday lives.

Farming, labour and trade

15The local farming environment, where small plots of land are distributed by household, is largely Bonan speaking, but apart from harvest time, when wider family members often return to help, this context will not include working aged men, who go away for labour (Ch. da gong 打工) for months in a year. In the latter case, while this brings the men out of a general Bonan-speaking environment, the fact that labour for work projects is regularly undertaken by syndicates of villagers still enables opportunities for the language to be utilised informally. During the slower winter months, many of the men return to the village, and during that season some villagers take the opportunity to expand or renovate their courtyard homes (Ch. siheyuan 四合院). The work is done on the basis of reciprocity, as relations and neighbours assist each other with the expectation that the favour will be returned in due course. These and other social contexts continue to provide strong Bonan-speaking environments within the community.

16In 2013 there was a small grocery shop in the centre of the village; however, the vast majority of purchases and trade took place in Dahejia Township, where the language of communication was Northwest Mandarin. In addition to those going away to work for periods of time, other villagers set up businesses in Dahejia or further afield, rarely returning to live in their houses or farm their fields. As more villagers move to the local and regional towns to live and work, the Bonan language is likely to become harder to maintain, even in a currently relatively vital environment such as Ganhetan.

Religion

Summary: Ganhetan

18Overall, the results show that the Bonan language is relatively well preserved in Ganhetan. It is being transmitted intergenerationally to a high degree, and it is in active use within the home and local environment. Furthermore, language attitudes are also very positive across all age groups, with 42 out of 62 respondents (68%) stating that they believed that Bonan would “continue to survive” and 58 respondents (94%) claiming that protecting the language was either a “highly important” or a “relatively important” mission. On the other hand, however, speaking domains outside the immediate home and village context show more of a mixed picture, and the regional dominance of Northwest Mandarin still presents a strong challenge to the language’s ongoing maintenance, particularly as villagers continue to move away for periods of time, or even permanently, to pursue educational or work opportunities.

Dadun (population = 2 100; total number of interviewees = 69)

19Although Bonan is spoken by all age groups in Dadun, and outside as well as inside the home, there is evidence of diminished use in the youngest generation: 13 out of 31 respondents (42%) under twenty-one years old stated that they “generally don’t speak it”, compared with 2 out of 18 (11%) in the over forty age bracket. Furthermore, only 20 out of 31 respondents (64%) from the youngest age group claimed to have an ability in Bonan been “OK” and “fluent”, compared with 17 out of 18 (94%) for the oldest group. Across all age groups, 65 out of a total of 69 interviewees (94%) claimed to have at least some grasp of Bonan, while 51 respondents (74%) were utilising the language at least in the home. Similar to Ganhetan, the Dadun villagers held highly positive attitudes towards the language: 38 out of a total of 69 respondents (55%) stated they believed that Bonan would “continue to survive”, while 67 (97%) regarded protecting the language as either a “highly important” or a “relatively important” mission.

Gaoli (population = 1 856; total number of interviewees = 52)

21Bonan is still spoken in Gaoli, both in the home and other domains, but there is clear evidence of decreasing speaker trends. While 17 out of 19 (89%) of those over forty claimed a level of Bonan between “OK” and “fluent”, the corresponding figures for the youngest age group was only 5 out of 13 (38%), while 8 out of 13 (62%) in the youngest bracket stated that they “generally don’t speak it”. Across all ages, 50 out of a total of 52 interviewees (96%) claimed to have some grasp of the language, and 38 respondents (73%) were utilising Bonan at least in the home. Language attitudes were again positive; while slightly more interviewees believed that Bonan would “slowly die out” (27, or 52%) than would “continue to survive” (25, or 48%), a total of 48 respondents (92%) claimed that protecting the Bonan language was either a “highly important” or a “relatively important” mission.

Xiaojia (Bonan population = 799 out of 2 793 residents; total number of interviewees = 48)

23Compared with the previous villages, in Xiaojia, there was a marked drop in both language use and ability across all ages, particularly the middle group, where 11 out of 16 respondents (69%) stated that they “generally don’t speak” Bonan, as did over half (7 out of 13) of those in the lower age bracket. While the older age bracket had the highest professed language ability, with nearly half (9 out of 19) respondents claiming that they were “fluent”, a similar proportion (8 out of 19, or 42%) stated that they couldn’t speak Bonan or that they could only “speak a little”. Of all those under forty, only 10 out of 29 respondents (34%) professed to have a Bonan ability of between “OK” and “fluent”, suggesting intergenerational disruption is well underway. Across all age groups, 33 out of a total of 48 interviewees (69%) professed some ability in the language, and 22 respondents (46%) were utilising the language at least in the home domain. Language attitudes were again positive, with 27 out of 48 interviewees (56%) believing that Bonan would “continue to survive” and a total of 44 respondents (92%) stating that protecting the language was either “highly important” or “relatively important”.

Mixed village, mixed marriages

Xietao (estimated Bonan population = 750; number of interviewees = 53)

25In Xietao the results show a relatively low level of Bonan language use even among the older residents. While just over half (10 out of 19) of respondents in the over-forty age group professed a Bonan ability of between “OK” and “fluent”, 12 out of 19 (63%) in that same category stated that they “generally don’t speak” the language. Across all age groups, 42 out of a total of 53 interviewees (79%) professed some level of ability, but only 16 out of 53 respondents (30%) were using the language at least in the home domain. Language attitudes were relatively positive, with 31 respondents (58%) believing that Bonan would “continue to survive” and a total of 43 respondents (81%) claiming that protecting the language was either “highly important” or “relatively important”.

26As a small population in a mixed village, the Xietao Bonan face a similar challenge in maintaining their language as do those from Xiaojia. In addition, while all the other villages and most other pockets of Bonan speakers are within the adjacent Dahejia and Liuji townships, Xietao is about a fourteen-kilometre drive away from Liuji, and located relatively close to Chuimatan (see fig. 2). Within such an isolated context, language shift towards Northwest Mandarin is already occurring at a rapid pace, and Bonan is in danger of eventually dying out in the village.

Summary of village results

27Overall, the results indicate clear disparities between villages. In Ganhetan, the Bonan language is still relatively well preserved across the generations, and utilised within the home and beyond. In Dadun and Gaoli, while the language is still being transmitted, and is also utilised outside the home, use and ability are decreasing generationally. In Xiaojia and Xietao, use is largely restricted to the home and there is evidence of advanced language shift. My results thus agree with Jian 2006 (against Satō 2006) in placing Ganhetan above Dadun in vitality, and I am also more optimistic than Satō about the status of Gansu Bonan across the five villages as a whole. Table 5 gives my amended version of Satō’s table, based on these results.

Table 5. Assessment of Gansu Bonan endangerment levels by village, based on the survey results, compared with Satō’s (2006) original assessment

Satō’s (2006) assessment		Revised assessment based on survey results
Danger Level	Bonan Village	Danger Level	Bonan Village
1. Relative to other villages, the influence of the Chinese language is not as strong.	Dadun Village 大墩村	1. Relative to other villages, the language is relatively well preserved, across generations and outside the home	Ganhetan Village 甘河滩村
2. Strongly influenced by the Chinese language.	Ganhetan Village 甘河滩村	2. The language is being transmitted intergenerationally and utilised outside the home, but use and ability are decreasing	Dadun Village 大墩村
3. Most people do not speak Bonan, or speak very simple Bonan.	Gaoli Village 高李村	3. Clear evidence of decreasing speaker trends across domains	Gaoli Village 高李村
4. Bonan is not spoken.	Xiaojia Village 肖家村 Xietao Village 斜套村	4. Most people under 40 don’t utilise the language regularly	Xiaojia Village 肖家村
		5. Most of the population don’t utilise the language regularly	Xietao Village 斜套村

Combined Results

29In terms of the specific endangerment scales themselves, firstly, the results concur with Ethnologue’s score of EGIDS level 6b, “Threatened: The language is used orally by all generations but only some of the child-bearing generation are transmitting it to their children”. However, Ethnologue currently applies this result to Bonan as a whole, rather than taking into account the different contexts of the two variants; Roche’s (2019a) analysis suggests the same level for Qinghai Bonan (although, similar to Gansu Bonan, with a high level of variation between villages), but further research may be required to confirm this. Gansu Bonan’s level on the Degree of Endangerment scale is harder to ascertain, partly because within UNESCO’s own documents there are slightly differing definitions of the various grades (table 6).

Table 6. The two scaling definitions used for Grade’s 3 and 4 in UNESCO’s Degree of Endangerment (adapted from Brenzinger et al. 2003, pp. 7-8, courtesy of the UNESCO; see also Moseley 2010, pp. 11-12; Norris 2010, p. 117)

Degree of Endangerment	Definition 1	Definition 2
Unsafe/Vulnerable (4)	Most but not all children or families of a particular community speak their language as their first language, but it may be restricted to specific social domains (such as at home where children interact with their parents and grandparents).	The language is used by some children in all domains; it is used by all children in limited domains.
Definitely Endangered (3)	The language is no longer being learned as the mother tongue by children in the home. The youngest speakers are thus of the parental generation. At this stage, parents may still speak their language to their children, but their children do not typically respond in the language.	The language is used mostly by the parental generation and up.

30Applying Definition 1, Bonan can be placed at Grade 4, “Unsafe/Vulnerable”, since most children still have at least some ability in the language, and it is not the case that the youngest speakers are of the parental generation (Grade 3, “Definitely Endangered”). This agrees with Lewis and Simons’ (2010) analysis that UNESCO Degree of Endangerment level 4 is the equivalent of their EGIDS level 6b (see table 1). But at the same time, it is unlikely in the Jishishan context that “the language is used by some children in all domains”, and given the speaker trends, it is now not being utilised by all children (Definition 2, Grade 4); rather, the language “is used mostly by the parental generation and up” (Definition 2, Grade 3), even if a good number are still passing it on to their children. As described earlier, UNESCO’s World Atlas of Languages places Bonan at Grade 3, “Definitely Endangered”, but the results of this research suggest a more nuanced level, if there was one, of “3,5”, in between “Vulnerable” and “Definitely Endangered”. In any case, this is a more optimistic result than Yang’s (2010) previous assessment of Gansu Bonan as lying between levels 3 (“Definitely Endangered”) and 2 (“Severely Endangered”), noted above. Finally, like Ethnologue, the World Atlas of Languages does not differentiate between Gansu and Qinghai Bonan.

31The LEI provides a relatively clearer picture. Based on this research, table 7 gives the individual grades for Gansu Bonan according to the four LEI factors, combined to make a percentage aggregate (with intergenerational transmission doubly weighted) of 32 percent; while table 8 shows the overall index, together with levels of certainty.

Table 7. Gansu Bonan’s Level of Endangerment, measured according to the four LEI factors. Scores for each factor range from 0 to 5, and intergenerational transmission is doubly weighted, giving a maximum possible score of 25. Since all four factors are used in the calculation, the level of certainty is considered to be “100% based on the evidence available”. For each factor’s individual gradation scale, see Appendix 2.

Factor	Score	LEI Definition	Weighted Score
Intergenerational transmission	1: Vulnerable	Most adults and some children are speakers	2
Absolute number of speakers	2: Threatened	1 000-9 999 speakers	2
Speaker number trends	2: Threatened	A majority of community members speak the language. Speaker numbers are gradually decreasing.	2
Domains of use	2: Threatened	Used in some nonofficial domains along with other languages, and remains the primary language in the home for many community members	2
Aggregate			8
Level of Endangerment, and aggregate score as a percentage	Threatened		8 out of a maximum possible score of 25 = 32%

32A 32 percent aggregate score places the language at the level of “Threatened”, and since all four factors are used in the calculation, the result is considered to be “100% certain, based on the evidence available” (see table 8). This result agrees with the LEI score on the ElCat website (currently at 40% certainty; see Catalogue of Endangered Languages 2023a); in addition, ElCat differentiates Gansu Bonan from Qinghai Bonan, which it also scores as “Threatened” (at 100% certainty; see Catalogue of Endangered Languages 2023b). However, the site puts the number of Gansu Bonan speakers at 10 000 to 15 000, which is a range well above the estimate from this present study; furthermore, the language is not restricted to just three villages, as the site also suggests.

33In summary, “Threatened”, the label given to both the relevant EGIDS and LEI levels, is an appropriate descriptive term for the current status of Gansu Bonan. While it is still spoken intergenerationally, and maintained to a reasonable level, particularly in Ganhetan, but also, to varying degrees, in Dadun and Gaoli, if no revitalisation efforts are undertaken, Northwest Mandarin is likely to continue to encroach at an accelerated pace, taking over all external domains, and eventually even the home domain itself.

Conclusion

34As has been alluded to earlier, in recent years, scholars have paid attention to the plight of cultural Tibet’s minority languages, including both those of official ethnicities, and those which are not recognised by the state (Roche & Suzuki 2018). By analysing the status of the Gansu Bonan language, this article highlights a related challenge facing the smaller minority Muslim communities of the Amdo region in relation to the dominant Sinophone Hui. As their languages continue to be encroached by Northwest Mandarin, their populations will become increasingly “Huiified”, resulting in a loss of diversity among the Muslims of Amdo. More work is required, both to test these provisional results for the Gansu Bonan, and to gain greater understanding with regards to the sociolinguistic situation of the other groups. It is hoped that this heuristic analysis provides a basis for further research on the non-Sinitic Muslim communities of the Amdo sprachbund.